Researchers Develop Method for Getting High-Quality Photos from Crappy Lenses

simplesample

There are many reason high-quality lenses cost as much as they do (and in some cases that is quite a lot), and one of them is that high-end lenses use many specially-designed elements that are perfectly-positioned to counteract aberrations and distortions.

But what if you could correct for all of that in post? Automatically? With just the click of a button? You could theoretically use a crappy lens and generate high-end results. Well, that’s what researchers at the University of British Columbia are working on, and so far their results are very promising.

The technique was presented at SIGGRAPH 2013, and it may some day provide a software alternative for those who can’t afford high-end glass. For their experiments, they developed a hand-made lens using only one element and then processed the resulting test images through their software to generate sharper results.

Check out their SIGGRAPH video below:

We won’t get into the technical bits (you can read the full paper here) but the basic premise is that once this software knows the point spread functions (PSFs) of your cheap lens, it can correct for blur, distortion and aberration and “recover” a high-quality image.

Here are some photos that show how their test images looked before (top) and after (bottom) sharpening with their computational imaging techniques:

simplesample1

simplesample2

simplesample3

simplesample4

simplesample5

simplesample6

simplesample7

simplesample8

The results are impressive, but for now there are still many hurdles left to jump before something like this could be brought to market. They have to figure out a way to account for the different PSFs of objects at different distances, and if the aperture gets any more open than f/2, the system runs into issues.

Still, this is a very promising start. To read more about the technique, check out the full paper. And if you’d like to see how their technique fared when using a non-homebrew lens — specifically, a Canon EF 28-105mm Macro — head over to this link for high-res samples.


Image credits: Photographs courtesy of the University of British Columbia

Discussion